Originally Posted by noah katz
If you want to complain about archaic technology, how about jet engines, which produce forward propulsion by throwing a crapload of greenhouse gases out of the back.
Well, even though you're moving to a non-sequitor here, I'll indulge you.
Modern jet engines don't work that way. You're describing the pure turbojet.
Modern jet engines, at least those used in sub-sonic aircraft, provide the vast majority of thrust from the ducted fan principle, thus are called turbofans. The turbine is used to supply mechanical energy to rotate the large fan and most of the thrust results from the fan, not the turbine exhaust. Modern turbofans have bypass ratios of 9:1 or better, some in commercial use up to 12:1.
--- Now getting back to topic-----
The premise of the conventional loudspeaker and amplifier is that the amplifier will build a very strong and perfectly accurate analog signal to feed to the voice coil(s) of the speaker, in the hopes that it will somehow turn this into a highly accurate audible representation of the original signal long ago, that impressed upon the microphone. Let's presume that the recording is what is intended to be heard, since we can't usually change the recording once it is made.
But let's look at the premise of the speaker/amplifier. It is a bit ludicrous to think that a mechanical system with all those resonances and unpredictable physical reactions can do this without any significant feedback or control loop.
Speakers are the big unknown in any playback system. They are the biggest source of distortion and coloration. Not only do they interact with their own components, the interact with the room and the objects in the room to a great extent.
If the ultimate goal of the machine is to vibrate the air as accurately as possible, wouldn't it make sense to use some sort of control loop in that process?
For instance, if you start with the standard coil/cone loudspeaker and measure its position and velocity and compare that to the intended function of the sound signal, you will find significant error. This goes far beyond simple corrections, like frequency response. Further, if you measure how that couples into the surrounding air, you'll find a great deal more error.
Now lets say you had a very powerful digital signal processor in the loop and it was programmed with all of the various transfer functions that accurately predicted what the speaker would mechanically do for any input condition and it could use the sensors integrated into the speaker to assure that it was doing just that, then the processor could take the original audio signal from the playback source, apply the necessary mathematical transformations and control calculations and cause the speaker to accurately excite the air with the signal.
You would have to expect that the current/voltage signal applied to the voice coil terminals, would look very, very different than what would come from a conventional amplifier. Yet, the net effect of the system would produce much more accurate vibrations into the air, which is the whole goal to begin with.
Such a system would allow for far more optimal design of the amplifier, speaker motors and enclosures than is possible from the conventional paradigm of "pretend it is perfect" and drive it hard.